I based on the documentation of PostgreSQL
(http://www.postgresql.org/docs/9.2/static/datatype-numeric.html#DATATYPE-FLOAT)which says about 'Infinity' and
'-Infinity'and doesn't mention other possible spellings, including 'inf'.
And on my installation of 9.2.4 'inf' doesn't work too (as I supposed according to documentation):
SELECT 'inf'::float8
ERROR: invalid input syntax for type double precision: "inf"
LINE 2: SELECT 'inf'::float8
According to Python's documentation (http://docs.python.org/2/library/stdtypes.html#typesnumeric), handle of infinities
andNaNs was added in 2.6. At least this works in 2.7.1:
Python 2.7.1 (r271:86832, Apr 1 2013, 01:27:27) [MSC v.1600 64 bit (AMD64)] onwin32
Type "help", "copyright", "credits" or "license" for more information.
>>> float('inf')
inf
>>> float('infinity')
inf
>>> float('+inf')
inf
>>> float('-inf')
-inf
So, Python is honest in this way. And, C99 says that style of representation ('inf' or 'infinity') is
implementation-defined,so it is all OK with modern Python.
> We already backstop strtod() for these cases:
>
> NaN
> Infinity
> -Infinity
>
> but the wording of the spec clearly requires +Infinity as well as the
> forms with just "inf". (It also appears to require +/- NaN to be
> accepted, but I have no idea what that would mean and suspect it to
> be a thinko.)
As I can judge, signed NaNs are from the same world as signed zeros and signed infinities. Strictly speaking: (-0)/(+0)
is-NaN, (-inf)/(+inf) is -NaN, and so on.
I think that PostgreSQL's ability to handle signed zeros (and all other rare stuff) depends on compiler used. Google
saysme that '-NaN' exists in modern glibc. I don't know about MSVC. My Python accepts '-nan' as input, but doesn't give
me'-NaN' as output.
So, I think it would be good if '-NaN' and other forms were workable.
--
Best regards,
Basil Peace
